In the last few years our group has pioneered an innovative minimally invasive approach to the heart for a growing portfolio of therapies that appeals to both minimally invasive cardiac surgeons and interventional specialists (e.g., cardiac electrophysiologists). The approach involves direct subxiphoid access to the pericardial space without interfering with the chest wall integrity and physiology. The clinical procedures that could benefit from such an approach include many that are performed wholly or primarily within the intrapericardial space. To date, many such procedures are performed thoracoscopically, necessitating differential lung ventilation and general endotracheal anesthesia. During the original period of this R01 grant, we have demonstrated the feasibility of intrapericardial intervention on the beating heart without entering the pleural space by developing [unreadable]HeartLander[unreadable], a tethered miniature robotic crawler that enters the pericardium via subxiphoid incision, attaches itself directly to the surface of the beating heart, moves to the desired epicardial location, and delivers therapy under the control of the surgeon. A prominent clinical goal in recent years in cardiac medicine has been the concept of combining diagnosis and treatment in a single session, sometimes referred to as [unreadable]one-stop shopping.[unreadable] HeartLander is well suited as a delivery vehicle for both diagnostic and interventional tools of various types, and as such is an ideal tool for realizing this vision. Potential clinical applications are numerous; as an example application we propose to focus the development of the technology on its relevance to heart failure. We hypothesize that HeartLander can autonomously map an infarct region epicardially and intervene (delineating the region with injected ink) more accurately and more rapidly than a trained clinician using state-of-the-art hand-guided tools. This research aims to develop methods for image-guided autonomous locomotion of HeartLander, first using static heart image data, and then using dynamic (beating) heart imagery. Techniques for image-guided autonomous epicardial mapping will be developed. Finally, methods will be developed for simultaneous image-guided autonomous cardiac mapping and intervention in a single session. All techniques developed will be evaluated, first in appropriate artificial heart phantoms, and then in a porcine model in vivo.